matlab图像处理-外文翻译-外文文献-英文文献-基于视觉的矿井救援.doc

《matlab图像处理-外文翻译-外文文献-英文文献-基于视觉的矿井救援.doc》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、附录A英文原文ScenerecognitionforminerescuerobotlocalizationbasedonvisionCUIYi-an(崔益安),CAIZi-xing(蔡自兴),WANGLu(王璐)Abstract：AnewscenerecognitionsystemwaspresentedbasedonfuzzylogicandhiddenMarkovmodel(HMM)thatcanbeappliedinminerescuerobotlocalizationduringemergenc

2、ies.Thesystemusesmonocularcameratoacquireomni-directionalimagesofthemineenvironmentwheretherobotlocates.Byadoptingcenter-surrounddifferencemethod,thesalientlocalimageregionsareextractedfromtheimagesasnaturallandmarks.TheselandmarksareorganizedbyusingHMMt

3、orepresentthescenewheretherobotis,andfuzzylogicstrategyisusedtomatchthesceneandlandmark.Bythisway,thelocalizationproblem,whichisthescenerecognitionprobleminthesystem,canbeconvertedintotheevaluationproblemofHMM.Thecontributionsoftheseskillsmakethesystemha

4、vetheabilitytodealwithchangesinscale,2Drotationandviewpoint.Theresultsofexperimentsalsoprovethatthesystemhashigherratioofrecognitionandlocalizationinbothstaticanddynamicmineenvironments.Keywords:robotlocation;scenerecognition;salientimage;matchingstrateg

5、y;fuzzylogic;hiddenMarkovmodel1IntroductionSearchandrescueindisasterareainthedomainofrobotisaburgeoningandchallengingsubject[1].Minerescuerobotwasdevelopedtoenterminesduringemergenciestolocatepossibleescaperoutesforthosetrappedinsideanddeterminewhetherit

6、issafeforhumantoenterornot.Localizationisafundamentalprobleminthisfield.Localizationmethodsbasedoncameracanbemainlyclassifiedintogeometric,topologicalorhybridones[2].Withitsfeasibilityandeffectiveness,scenerecognitionbecomesoneoftheimportanttechnologieso

7、ftopologicallocalization.Currentlymostscenerecognitionmethodsarebasedonglobalimagefeaturesandhavetwodistinctstages:trainingofflineandmatchingonline.Duringthetrainingstage,robotcollectstheimagesoftheenvironmentwhereitworksandprocessestheimagestoextractglo

8、balfeaturesthatrepresentthescene.Someapproacheswereusedtoanalyzethedata-setofimagedirectlyandsomeprimaryfeatureswerefound,suchasthePCAmethod[3].However,thePCAmethodisnoteffectiveindistinguishingtheclassesoffeatures.Another